Research & Innovation
Problems we're working on.
This is applied research, not an academic programme. Each area below is something we hit in delivery, could not solve with what existed, and are still working on.
01 — Active areas
Five open questions.
We publish what we learn where it is useful and admit where we are still guessing. None of these is solved.
- 01
Assessment integrity
How to make remote assessment defensible rather than merely surveilled — measurement design, similarity analysis, and the review process around automated flags.
- 02
Grounded educational AI
Constraining assistants so they explain without completing. Retrieval grounding, refusal behaviour, and measuring whether an assistant improved learning or replaced it.
- 03
Readiness modelling
Composite scoring that predicts hiring outcomes better than marks do, and the fairness questions that come with any such model.
- 04
Low-bandwidth delivery
Learning and assessment software that degrades gracefully on intermittent connections — local-first patterns, conflict resolution, sync.
- 05
Skills-demand signalling
Extracting regional skills demand from hiring activity, and feeding it back into curriculum design faster than the accreditation cycle allows.
02 — How we work on this
Research that has to survive Tuesday.
Everything here is tested against live delivery within a term. If an approach to readiness scoring cannot be defended to a head of department who disagrees with its output, it is not finished.
That constraint rules out a lot of interesting work. It also means what remains is usable, and that we find out quickly when we are wrong — which happens often enough to be worth designing for.
03 — Publications
Written output.
Nothing formally published yet. Our practitioner writing is on the blog, which is where the current thinking actually lives.
Read the blogNo peer-reviewed publications or formal white papers to date. We would rather show an empty section than pad it. When there is something worth citing it will appear here with a link to the original.
1 placeholder in the content registry
Working on the same problems?
If you are researching assessment integrity, readiness prediction or educational AI, we have data and delivery access, and we are interested in collaborating.